基于MPI和OpenMP混合编程的高分三号数据分布式并行转换算法

高分三号是我国C波段多极化合成孔径雷达卫星。PolSARpro是欧空局支持下的一款极化SAR影像处理的开源软件,为了便于利用该软件处理高分三号数据,本文提出了一种基于MPI和OpenMP并以PolSARpro软件的数据格式要求进行分布式并行转换算法,实现将高分三号极化数据快速精确转化为复数散射矩阵S2数据格式,通过KingMap V8.0平台实现了算法并在实际数据中进行测试,验证了算法的可行性、正确性和高效性。

基于CGA的MPI程序分支覆盖测试套件生成

针对程序的分支覆盖测试,元启发式搜索技术已经被广泛应用于测试数据生成中。然而,当前的研究成果主要适用于串行程序。因此,为覆盖消息传递接口(Message Passing Interface,MPI)程序的分支,该文研究基于协同进化遗传算法(Co-evolutionary Genetic Algorithm,CGA)的测试套件生成方法(简称为:CGA生成法),该方法具有不受不可行分支影响的优势。首先,基于收集覆盖信息的探针,定义最小归一化分支距离,并以此设计出相应的适应度值函数;然后,使用CGA生成进化个体,并基于设计的适应度值函数,计算这些个体的适应值;最后,基于计算的适应值,选择子种群中代表个体,以构成合作种群。所提CGA生成法应用于7个基准MPI程序,并与其他多种方法进行比较。实验结果表明,CGA生成法的覆盖率通常高于其他搜索算法。

基于MPI+CUDA的DSMC/PIC耦合模拟异构并行及性能优化研究

DSMC/PIC耦合模拟是一类重要的高性能计算应用,大规模DSMC/PIC耦合模拟计算量巨大,需要实现高效并行计算。由于粒子动态注入、迁移等操作,基于MPI并行的DSMC/PIC耦合模拟往往通信开销较大且难以实现负载均衡。针对自主研发的DSMC/PIC耦合模拟软件,在原有MPI并行优化版本上设计实现了高效的MPI+CUDA异构并行算法,结合GPU体系结构和DSMC/PIC计算特点,开展了GPU访存优化、GPU线程工作负载优化、CPU-GPU数据传输优化及DSMC/PIC数据冲突优化等一系列性能优化。在北京北龙超级云HPC系统的NVIDIA V100和A100 GPU上,针对数亿粒子规模的脉冲真空弧等离子体羽流应用,开展了大规模DSMC/PIC耦合异构并行模拟,相比原有纯MPI并行,GPU异构并行大幅缩短了模拟时间,两块GPU卡较192核的CPU加速比达到550%,同时具有更好的强可扩展性。

利用MPI实现点云SAC-IA并行配准

采样一致性初始配准算法(SAC-IA)是点云的一种粗配准算法。针对大规模点云SAC-IA配准效率低、实时性差等问题,本文提出利用消息传递接口MPI实现点云SAC-IA多进程并行配准,主要包括法向量并行估计、SPFH特征及FPFH特征并行计算和SAC-IA并行配准。实验结果表明,MPI多进程并行算法可显著提高点云SAC-IA配准速度。

一种基于HDFS的分布式文件系统MPIFS

传统的MPI(Message Passing Interface)计算特点是数据向计算迁移,对于数据量庞大的计算任务具有先天的不足。文中提出一种支持MPI的分布式文件系统MPIFS的架构及实现。该文件系统基于HDFS(Hadoop Distributed File System),使得MPI在MPIFS上能同时支持计算密集型和数据密集型计算,设置两个类型的批处理词频统计实验,所需数据都分布式存储在MPIFS分布式文件系统中,通过调用系统提供的统一数据接口实现数据访问。1个计算节点在本地计算大小为m的文件,n个计算节点分布式并行计算大小为n×m的文件,两者计算时间相同,MPIFS中文件总量不变,计算节点数量减少,计算时间t变长,可得出MPIFS文件系统架构可行,能够支持MPI实现计算向数据迁移的并行计算。

压电陶瓷作动器的MPI动态迟滞建模与控制

压电陶瓷是一种具有迟滞非线性的智能材料。为了实现系统的精密跟踪控制,提出一种基于MPI(modified Prandtl-Ishlinskii)的Hammerstein动态迟滞模型,并基于该模型设计了滑模跟踪控制方案。在play算子的上升边沿和下降边沿阈值处引入了延时系数,并串联死区算子构成改进的非对称PI(Prandtl-Ishlinskii)模型,基于MPI的Hammerstein动态迟滞非线性模型可以描述压电陶瓷作动器的率相关迟滞特性。通过采集在单频率10 Hz,40 Hz,80 Hz和复合频率10~90 Hz正弦输入电压信号下的压电陶瓷作动器的位移数据,并采用粒子群算法和最小二乘递推方法辨识MPI模型参数和ARX(auto regressive model with exogenous input)模型参数,验证了模型的可行性,相较于基于经典PI的Hammerstein动态迟滞模型,模型误差分别降低了37%,42%,35%和24%。最后,构建迟滞补偿器,利用Hammerstein模型的模块化特点,提出一种可以实现对系统动态跟踪控制的滑模控制方案,并搭建了滑模控制压电系统试验平台,对单频率1 Hz,40 Hz,80 Hz和复合频率10~90 Hz的正弦输入电压信号进行了微位移实时跟踪控制试验,试验中的相对误差在7.62%以内,均方根最大误差为1.8573μm,表明所提出的滑模控制器有较强的跟踪性能。

Ionization Engineering of Hydrogels Enables Highly Efficient Salt‑Impeded Solar Evaporation and Night‑Time Electricity Harvesting

Interfacial solar evaporation holds immense potential for brine desalination with low carbon footprints and high energy utilization.Hydrogels,as a tunable material platform from the molecular level to the macroscopic scale,have been considered the most promising candidate for solar evaporation.However,the simultaneous achievement of high evaporation efficiency and satisfactory tolerance to salt ions in brine remains a challenging scientific bottleneck,restricting the widespread application.Herein,we report ionization engineering,which endows polymer chains of hydrogels with electronegativity for impeding salt ions and activating water molecules,fundamentally overcoming the hydrogel salt-impeded challenge and dramatically expediting water evaporating in brine.The sodium dodecyl benzene sulfonate-modified carbon black is chosen as the solar absorbers.The hydrogel reaches a ground-breaking evaporation rate of 2.9 kg m−2 h−1 in 20 wt%brine with 95.6%efficiency under one sun irradiation,surpassing most of the reported literature.More notably,such a hydrogel-based evaporator enables extracting clean water from oversaturated salt solutions and maintains durability under different high-strength deformation or a 15-day continuous operation.Meantime,on the basis of the cation selectivity induced by the electronegativity,we first propose an all-day system that evaporates during the day and generates salinity-gradient electricity using waste-evaporated brine at night,anticipating pioneer a new opportunity for all-day resource-generating systems in fields of freshwater and electricity.

基于MPI的鲲鹏CPU核间通信研究

核间通信延时是影响高性能计算系统整体运行效率的重要因素.国产鲲鹏CPU在高性能计算领域应用日益广泛,针对鲲鹏CPU的缓存架构及多核间接口互联进行分析,研究影响鲲鹏CPU核间通信延时的因素.在消息传递接口(MPI)环境下进行节点内核间通信实验,对包括跨三级缓存、跨物理CPU通信等不同模式下通信延时进行对比,发现通信数据包大于500 KB后,跨L3 Cache TAG的通信延时反优于共享L3 Cache TAG的通信延时.针对通信数据包在64 KB大小时的通信延迟异常,分析得出是MPI的Eager模式和Rendezvous模式的默认切换阈值所造成.对这两种模式进行实验对比,验证不同大小的通信数据包在不同模式下和跨核通信时的延时特征,Eager模式更适合低延时的小消息发送.在实际应用中可根据通信数据包大小调整两种模式的默认切换阈值,以达到更好的传输效果.实验结果表明由于鲲鹏CPU存在复杂的多核结构,在并行计算程序设计时可以进行针对性优化,以提升程序的运行效率.

An MPI parallel DEM-IMB-LBM framework for simulating fluid-solid interaction problems

The high-resolution DEM-IMB-LBM model can accurately describe pore-scale fluid-solid interactions,but its potential for use in geotechnical engineering analysis has not been fully unleashed due to its prohibitive computational costs.To overcome this limitation,a message passing interface(MPI)parallel DEM-IMB-LBM framework is proposed aimed at enhancing computation efficiency.This framework utilises a static domain decomposition scheme,with the entire computation domain being decomposed into multiple subdomains according to predefined processors.A detailed parallel strategy is employed for both contact detection and hydrodynamic force calculation.In particular,a particle ID re-numbering scheme is proposed to handle particle transitions across sub-domain interfaces.Two benchmarks are conducted to validate the accuracy and overall performance of the proposed framework.Subsequently,the framework is applied to simulate scenarios involving multi-particle sedimentation and submarine landslides.The numerical examples effectively demonstrate the robustness and applicability of the MPI parallel DEM-IMB-LBM framework.

MPI+CUDA联合加速重力场反演的并行算法

针对重力场解算过程中数据量巨大的问题,联合MPI(massage passing interface)与CUDA(compute unified device architecture)提出基于最小二乘法的重力场解算过程的并行加速算法。使用MPI完成复杂过程的任务分配,实现全局层面的并行加速;基于CUDA编写大规模矩阵相乘的并行加速程序,并针对不同类型的矩阵进行适配,同时联合MPI将法矩阵的计算过程进一步细分,实现对分进程内存峰值的压缩。在单机上完成30阶与120阶重力场仿真解算任务,结果表明,反演30阶重力场时加速比可达180;反演120阶重力场时,并行计算单次迭代仅耗时2 h,而串行模式下无法计算。

MPI/OpenMP-Based Parallel Solver for Imprint Forming Simulation

In this research,we present the pure open multi-processing(OpenMP),pure message passing interface(MPI),and hybrid MPI/OpenMP parallel solvers within the dynamic explicit central difference algorithm for the coining process to address the challenge of capturing fine relief features of approximately 50 microns.Achieving such precision demands the utilization of at least 7 million tetrahedron elements,surpassing the capabilities of traditional serial programs previously developed.To mitigate data races when calculating internal forces,intermediate arrays are introduced within the OpenMP directive.This helps ensure proper synchronization and avoid conflicts during parallel execution.Additionally,in the MPI implementation,the coins are partitioned into the desired number of regions.This division allows for efficient distribution of computational tasks across multiple processes.Numerical simulation examples are conducted to compare the three solvers with serial programs,evaluating correctness,acceleration ratio,and parallel efficiency.The results reveal a relative error of approximately 0.3%in forming force among the parallel and serial solvers,while the predicted insufficient material zones align with experimental observations.Additionally,speedup ratio and parallel efficiency are assessed for the coining process simulation.The pureMPI parallel solver achieves a maximum acceleration of 9.5 on a single computer(utilizing 12 cores)and the hybrid solver exhibits a speedup ratio of 136 in a cluster(using 6 compute nodes and 12 cores per compute node),showing the strong scalability of the hybrid MPI/OpenMP programming model.This approach effectively meets the simulation requirements for commemorative coins with intricate relief patterns.

计及需求响应的MPIES多时空尺度鲁棒随机优化调度

针对多园区综合能源系统多时间尺度优化调度问题,结合各时间尺度源、荷不确定性特征与响应特性、园区数目对计算规模的影响,提出一种计及需求响应的多园区综合能源系统多时空尺度鲁棒随机优化调度架构。日前调度建立计及可中断负荷的多园区多能流系统鲁棒优化调度模型,确定系统最恶劣运行场景下基本调度策略;日内调度建立计及可中断负荷和替代型需求响应的单园区多能流系统随机优化调度模型,细化并调整调度策略;实时调度建立计及直接负荷控制的单园区电气系统随机优化调度模型,削弱调度策略受源、荷出力波动的影响。最后,通过算例验证了该架构能有效提高系统运行的经济性和稳定性。

Theoretical investigation of electron-impact ionization of W8+ion

The electron-impact single ionization cross section for W8+ion has been calculated using flexible atomic code,employing the level-to-level distorted-wave approximation.This calculations takes into account contributions form both direct ionization(DI)and excitation autoionization(EA).However,the theoretical predictions,based solely on the ground state,tends to underestimate the experimental values.This discrepancy can be mitigated by incorporation contributions from excited states.We extended the theoretical analysis,including the contributions from the long-lived metastable states with lifetimes exceeding 1.5×10-5 s.We employed two statistical models to predict the fraction of ground state ions in the parent ion beam.Assuming a 79%fraction of parent ions in ground configuration,the experiment measurements align with the predictions.Furthermore we derived the theoretical cross-section for the ground state as correlated plasma rate coefficients,and compared it with existing data.Despite the uncertainty in our calculations,our results are still acceptable.

Polarization control of above-threshold ionization spectrum in elliptically polarized two-color laser fields

We study the above-threshold ionization(ATI)process of atoms exposed to fundamental and high-frequency lasers with arbitrary ellipticity by applying the frequency-domain theory.It is found that the angular-resolved ATI spectrum is sensitive to ellipticities of two lasers and emitted angles of the photoelectron.Particularly for the photon energy of the highfrequency laser more than atomic ionization potential,the width of plateau tends to a constant with increasing ellipticity of fundamental field,the dip structure disappears with increasing ellipticity of the high-frequency field.With the help of the quantum channel analysis,it is shown that the angular distribution depends mainly on the ellipticity of high-frequency field in the case that its frequency is high.Moreover,one can see that the maximal and minimal energies in quantum numerical results are in good agreement with the classical prediction.Our investigation may provide theoretical support for experimental research on polarization control of ionization in elliptically polarized two-color laser fields.

Excitation and ionization of OCS molecules in strong UV and NIR laser fields

Rydberg state excitation(RSE) is a highly non-linear physical phenomenon that is induced by the ionization of atoms or molecules in strong femtosecond laser fields. Here we observe that both parent and fragments(S, C, OC) of the triatomic molecule carbonyl sulfide(OCS) can survive strong 800 nm or 400 nm laser fields in high Rydberg states. The dependence of parent and fragment RSE yields on laser intensity and ellipticity is investigated in both laser fields, and the results are compared with those for strong-field ionization. Distinctly different tendencies for laser intensity and ellipticity are observed for fragment RSE compared with the corresponding ions. The mechanisms of RSE and strong-field ionization of OCS molecules in different laser fields are discussed based on the experimental results. Our study sheds some light on the strong-field excitation and ionization of molecules irradiated by femtosecond NIR and UV laser fields.

Ion acoustic solitary waves in an adiabatic dusty plasma:Roles of superthermal electrons,ion loss and ionization

We investigate propagation of dust ion acoustic solitary wave(DIASW)in a multicomponent dusty plasma with adiabatic ions,superthermal electrons,and stationary dust.The reductive perturbation method is employed to derive the damped Korteweg-de Vries(DKdV)equation which describes DIASW.The result reveals that the adiabaticity of ions significantly modifies the basic features of the DIASW.The ionization effect makes the solitary wave grow,while collisions reduce the growth rate and even lead to the damping.With the increases in ionization cross sectionΔσ/σ_(0),ion-to-electron density ratioδ_(ie)and superthermal electrons parameterκ,the effect of ionization on DIASW enhances.

Internal collision double ionization of molecules driven by co-rotating two-color circularly polarized laser pulses

The double ionization process of molecules driven by co-rotating two-color circularly polarized fields is investigated with a three-dimensional classical ensemble model. Numerical results indicate that a considerable part of the sequential double ionization(DI) events of molecules occur through internal collision double ionization(ICD), and the ICD recollision mechanism is significantly different from that in non-sequential double ionization(NSDI). By analyzing the results of internuclear distances R = 5 a.u. and 2 a.u., these two recollision mechanisms are studied in depth. It is found that the dynamic behaviors of the recollision mechanisms of NSDI and ICD are similar. For NSDI, the motion range of electrons after the ionization is relatively large, and the electrons will return to the core after a period of time. In the ICD process,electrons will rotate around the parent ion before ionization, and the distance of the electron motion is relatively small. After a period of time, the electrons will come back to the core and collide with another electron. Furthermore, the molecular internuclear distance has a significant effect on the electron dynamic behavior of the two ionization mechanisms. This study will help to understand the multi-electron ionization process of complex systems.

Manipulating the electron dynamics in the non-sequential double ionization process of Ar atoms by an orthogonal two-color laser field

Electron dynamics during non-sequential double ionization(NSDI) is one of the most attractive areas of research in the field of laser–atom or laser–molecule interaction. Based on the classic two-dimensional model, we study the process of NSDI of argon atoms driven by a few-cycle orthogonal two-color laser field composed of 800 nm and 400 nm laser pulses. By changing the relative phase of the two laser pulses, a localized enhancement of NSDI yield is observed at 0.5πand 1.5π, which could be attributed to a rapid and substantial increase in the number of electrons returning to the parent ion within extremely short time intervals at these specific phases. Through the analysis of the electron–electron momentum correlations within different time windows of NSDI events and the angular distributions of emitted electrons in different channels, we observe a more pronounced electron–electron correlation phenomenon in the recollision-induced ionization(RII) channel. This is attributed to the shorter delay time in the RII channel.

Absolute partial and total ionization cross sections of carbon monoxide with electron collision from 350 eV to 8000 eV

The absolute partial and total cross sections for electron impact ionization of carbon monoxide are reported for electron energies from 350 eV to 8000 eV.The product ions(CO^(+),C^(+),O^(+),CO^(2+),C^(2+),and O^(2+))are measured by employing an ion imaging mass spectrometer and two ion-pair dissociation channels(C^(+)+O^(+)and C^(2+)+O^(+))are identified.The absolute cross sections for producing individual ions and their total,as well as for the ion-pair dissociation channels are obtained by normalizing the data of CO^(+)to that of Ar^(+)from CO-Ar mixture target with a fixed 1:1 ratio.The overall errors are evaluated by considering various kinds of uncertainties.A comprehensive comparison is made with the available data,which shows a good agreement with each other over the energy ranges that are overlapped.This work presents new cross-section data with electron energies above 1000 eV.

First Born Double Differential Cross-Section for Ionization of H (3d) by Incident Electron Impact

Double differential cross section (DDCS) of First-Born approximation is calcu-lated for the ionization of metastable 3d-state hydrogen atoms by electron impact energy at 150 eV and 250 eV. A multiple scattering theory is applied in the present study. The present results are compared with the other related the-oretical results for the ionization of hydrogen atoms from different metastable states and ground-state experimental results. The findings demonstrate a strong qualitative agreement with the existing results. The obtained results have an extensive scope for further study of such an ionization process.